NAACL 2024 论文列表

Large Language Models (LLMs) are powerful models for generation tasks, but they may not generate good quality outputs in their first attempt. Apart from model fine-tuning, existing approaches to improve prediction accuracy and quality typically involve LLM self-improvement / self-reflection that incorporate feedback from models themselves. Despite their effectiveness, these methods are hindered by their high computational cost and lack of scalability. In this work, we propose CERET, a method for refining text generations by considering semantic stability, entailment and inter-sample uncertainty measures. Experimental results show that CERET outperforms Self-consistency and Self-rerank baselines consistently under various task setups, by 1.6% in Rouge-1 for abstractive summarization and 3.5% in hit rate for question answering. Compared to LLM Self-rerank method, our approach only requires 9.4% of its latency and is more cost-effective.

In text ranking, it is generally believed that the cross-encoders already gather sufficient token interaction information via the attention mechanism in the hidden layers. However, our results show that the cross-encoders can consistently benefit from additional token interaction in the similarity computation at the last layer. We introduce CELI (Cross-Encoder with Late Interaction), which incorporates a late interaction layer into the current cross-encoder models. This simple method brings 5% improvement on BEIR without compromising in-domain effectiveness or search latency. Extensive experiments show that this finding is consistent across different sizes of the cross-encoder models and the first-stage retrievers. Our findings suggest that boiling all information into the [CLS] token is a suboptimal use for cross-encoders, and advocate further studies to investigate its relevance score mechanism.

Training a supervised news summarization model requires large amounts of high-quality training data consisting of news articles paired with reference summaries. However, obtaining such data is costly, and existing datasets contain considerable amount of noise. We present a new large-scale and high-quality dataset for supervised abstractive news summarization containing 1.3 million training samples, which we call CCSum. In creating this dataset, we take advantage of the journalistic inverted-pyramid style in news writing: In some articles, the first sentence can be considered a summary of the reported story. Accordingly, among 35 million CommonCrawl News articles, we identify pairs of articles about the same news story and use one article’s first sentence as the summary for the other article. To ensure high quality, we apply strict filters whose parameters we optimize using Bayesian optimization. We show that the resulting dataset is more factual and informative than established summarization datasets; less than 1% of the summaries have major factual inconsistencies with the corresponding news articles, compared to 5.5% to 15.4% in existing datasets, according to our human evaluation. Summarization models trained on our dataset are more favored compared to those trained on CNN/Daily Mail. The proposed dataset can open new opportunities for future research in abstractive summarization.

The argument sufficiency assessment task aims to determine if the premises of a given argument support its conclusion.To tackle this task, existing works often train a classifier on data annotated by humans. However, annotating data is laborious, and annotations are often inconsistent due to subjective criteria. Motivated by the definition of probability of sufficiency (PS) in the causal literature, we proposeCASA, a zero-shot causality-driven argument sufficiency assessment framework. PS measures how likely introducing the premise event would lead to the conclusion when both the premise and conclusion events are absent. To estimate this probability, we propose to use large language models (LLMs) to generate contexts that are inconsistent with the premise and conclusion and revise them by injecting the premise event.Experiments on two logical fallacy detection datasets demonstrate that CASA accurately identifies insufficient arguments. We further deploy CASA in a writing assistance application, and find that suggestions generated by CASA enhance the sufficiency of student-written arguments. Code and data are available at https://github.com/xxxiaol/CASA.

Extracting fine-grained experimental findings from literature can provide dramatic utility for scientific applications. Prior work has developed annotation schemas and datasets for limited aspects of this problem, failing to capture the real-world complexity and nuance required. Focusing on biomedicine, this work presents CARE—a new IE dataset for the task of extracting clinical findings. We develop a new annotation schema capturing fine-grained findings as n-ary relations between entities and attributes, which unifies phenomena challenging for current IE systems such as discontinuous entity spans, nested relations, variable arity n-ary relations and numeric results in a single schema. We collect extensive annotations for 700 abstracts from two sources: clinical trials and case reports. We also demonstrate the generalizability of our schema to the computer science and materials science domains. We benchmark state-of-the-art IE systems on CARE, showing that even models such as GPT4 struggle. We release our resources to advance research on extracting and aggregating literature findings.

Cultural variation exists between nations (e.g., the United States vs. China), but also within regions (e.g., California vs. Texas, Los Angeles vs. San Francisco). Measuring this regional cultural variation can illuminate how and why people think and behave differently. Historically, it has been difficult to computationally model cultural variation due to a lack of training data and scalability constraints. In this work, we introduce a new research problem for the NLP community: How do we measure variation in cultural constructs across regions using language? We then provide a scalable solution: building knowledge-guided lexica to model cultural variation, encouraging future work at the intersection of NLP and cultural understanding. We also highlight modern LLMs’ failure to measure cultural variation or generate culturally varied language.

We present BYOKG, a universal question-answering (QA) system that can operate on any knowledge graph (KG), requires no human-annotated training data, and can be ready to use within a day—attributes that are out-of-scope for current KGQA systems. BYOKG draws inspiration from the remarkable ability of humans to comprehend information present in an unseen KG through exploration—starting at random nodes, inspecting the labels of adjacent nodes and edges, and combining them with their prior world knowledge. Exploration in BYOKG leverages an LLM-backed symbolic agent that generates a diverse set of query-program exemplars, which are then used to ground a retrieval-augmented reasoning procedure to synthesize programs for arbitrary questions. BYOKG is effective over both small- and large-scale graphs, showing dramatic gains in zero-shot QA accuracy of 27.89 and 59.88 F1 on GrailQA and MetaQA, respectively. We further find that performance of BYOKG reliably improves with continued exploration as well as improvements in the base LLM, notably outperforming a state-of-the-art fine-tuned model by 7.08 F1 on a sub-sampled zero-shot split of GrailQA. Lastly, we verify our universality claim by evaluating BYOKG on a domain-specific materials science KG and show that it improves zero-shot performance by 46.33 F1.

Scaling high-quality tutoring remains a major challenge in education. Due to growing demand, many platforms employ novice tutors who, unlike experienced educators, struggle to address student mistakes and thus fail to seize prime learning opportunities. Our work explores the potential of large language models (LLMs) to close the novice-expert knowledge gap in remediating math mistakes. We contribute Bridge, a method that uses cognitive task analysis to translate an expert’s latent thought process into a decision-making model for remediation. This involves an expert identifying (A) the student’s error, (B) a remediation strategy, and (C) their intention before generating a response. We construct a dataset of 700 real tutoring conversations, annotated by experts with their decisions. We evaluate state-of-the-art LLMs on our dataset and find that the expert’s decision-making model is critical for LLMs to close the gap: responses from GPT4 with expert decisions (e.g., “simplify the problem”) are +76% more preferred than without. Additionally, context-sensitive decisions are critical to closing pedagogical gaps: random decisions decrease GPT4’s response quality by -97% than expert decisions. Our work shows the potential of embedding expert thought processes in LLM generations to enhance their capability to bridge novice-expert knowledge gaps. Our dataset and code can be found at: https://github.com/rosewang2008/bridge.

Ensembling different large language models (LLMs) to unleash their complementary potential and harness their individual strengths is highly valuable. Nevertheless, vocabulary discrepancies among various LLMs have constrained previous studies to either selecting or blending completely generated outputs. This limitation hinders the dynamic correction and enhancement of outputs during the generation process, resulting in a limited capacity for effective ensemble. To address this issue, we propose a novel method to Ensemble LLMs via Vocabulary Alignment (EVA). EVA bridges the lexical gap among various LLMs, enabling meticulous ensemble at each generation step. Specifically, we first learn mappings between the vocabularies of different LLMs with the assistance of overlapping tokens. Subsequently, these mappings are employed to project output distributions of LLMs into a unified space, facilitating a fine-grained ensemble. Finally, we design a filtering strategy to exclude models that generate unfaithful tokens. Experimental results on commonsense reasoning, arithmetic reasoning, machine translation, and data-to-text generation tasks demonstrate the superiority of our approach compared with individual LLMs and previous ensemble methods conducted on complete outputs. Further analyses confirm that our approach can leverage knowledge from different language models and yield consistent improvement.

Large language models hold significant promise in multilingual applications. However, inherent biases stemming from predominantly English-centric pre-training have led to the widespread practice of pre-translation, i.e., translating non-English inputs to English before inference, leading to complexity and information loss. This study re-evaluates the need for pre-translation in the context of PaLM2 models, which have been established as highly performant in multilingual tasks. We offer a comprehensive investigation across 108 languages and 6 diverse benchmarks, including open-end generative tasks, which were excluded from previous similar studies. Our findings challenge the pre-translation paradigm established in prior research, highlighting the advantages of direct inference in PaLM2. Specifically, PaLM2-L consistently outperforms pre-translation in 94 out of 108 languages. These findings pave the way for more efficient and effective multilingual applications, alleviating the limitations associated with pre-translation and unlocking linguistic authenticity.

Large Language Models (LLMs) are frequently used for multi-faceted language generation and evaluation tasks that involve satisfying intricate user constraints or taking into account multiple aspects and criteria. However, their performance can fall short, due to the model’s lack of coherence and inability to plan and decompose the problem. We propose Branch-Solve-Merge (BSM), a Large Language Model program (Schlag et al., 2023) for tackling such challenging natural language tasks. It consists of branch, solve, and merge modules that are parameterized with specific prompts to the base LLM. These three modules plan a decomposition of the task into multiple parallel sub-tasks, independently solve them, and fuse the solutions to the sub-tasks. We apply our method to the tasks of LLM response evaluation and constrained text generation and evaluate its effectiveness with multiple LLMs, including Vicuna, LLaMA-2-chat, and GPT-4. BSM improves the evaluation correctness and consistency for each LLM by enhancing human-LLM agreement by up to 26%, reducing length and pairwise position biases by up to 50%, and allowing LLaMA-2-chat to match or outperform GPT-4 on most domains. On a constraint story generation task, BSM improves the coherence of stories while also improving constraint satisfaction by 12%.

In the realm of modern Large Language Models (LLMs), facilitating high-quality, multi-turn dialogues with humans represents a cornerstone feature. However, human-based evaluation of such a capability involves substantial manual effort. This study offers a formative assessment of current LLMs’ proficiency in emulating human-like, multi-turn conversations using an LLM-centric approach. The evaluation encompasses three key elements in the evaluation pipeline: utterance generation, evaluation protocol, and judgement, and we delve deeply into each aspect. GPT-4, both as an utterance generator and as a judge, exhibits exceptional performance. As a generator, GPT-4 crafts dialogues indistinguishable from human interactions in terms of style and flow. When judging, it shows a heightened alignment with human evaluative standards and consistency. Conversely, other LLMs face challenges in producing quality multi-turn dialogues, hindered by inadequate instruction-following abilities, a propensity for prolix utterances, and overall limited capabilities. Notably, generating extensive dialogues (e.g., spanning tens of turns) remains a formidable task for most LLMs, particularly in Chinese contexts. We hope that our work can serve as a valuable resource for evaluating the multi-turn chatting capabilities of LLMs. Related resources are available at https://github.com/open-compass/BotChat.

Several large-scale datasets (e.g., WikiSQL, Spider) for developing natural language interfaces to databases have recently been proposed. These datasets cover a wide breadth of domains but fall short on some essential domains, such as finance and accounting. Given that accounting databases are used worldwide, particularly by non-technical people, there is an imminent need to develop models that could help extract information from accounting databases via natural language queries. In this resource paper, we aim to fill this gap by proposing a new large-scale Text-to-SQL dataset for the accounting and financial domain: BookSQL. The dataset consists of 100k natural language queries-SQL pairs, and accounting databases of 1 million records. We experiment with and analyze existing state-of-the-art models (including GPT-4) for the Text-to-SQL task on BookSQL. We find significant performance gaps, thus pointing towards developing more focused models for this domain.

Modern biomedical concept representations are mostly trained on synonymous concept names from a biomedical knowledge base, ignoring the inter-concept interactions and a concept’s local neighborhood in a knowledge base graph. In this paper, we introduce Biomedical Entity Representation with a Graph-Augmented Multi-Objective Transformer (BERGAMOT), which adopts the power of pre-trained language models (LMs) and graph neural networks to capture both inter-concept and intra-concept interactions from the multilingual UMLS graph. To obtain fine-grained graph representations, we introduce two additional graph-based objectives: (i) a node-level contrastive objective and (ii) the Deep Graph Infomax (DGI) loss, which maximizes the mutual information between a local subgraph and a high-level graph summary. We apply contrastive loss on textual and graph representations to make them less sensitive to surface forms and enable intermodal knowledge exchange. BERGAMOT achieves state-of-the-art results in zero-shot entity linking without task-specific supervision on 4 of 5 languages of the Mantra corpus and on 8 of 10 languages of the XL-BEL benchmark.

The pre-trained language model (PLM) has achieved significant success in the field of knowledge graph completion (KGC) by effectively modeling entity and relation descriptions. In recent studies, the research in this field has been categorized into methods based on word matching and sentence matching, with the former significantly lags behind. However, there is a critical issue in word matching methods, which is that these methods fail to obtain satisfactory single embedding representations for entities.To address this issue and enhance entity representation, we propose the Bilateral Masking with prompt for Knowledge Graph Completion (BMKGC) approach.Our methodology employs prompts to narrow the distance between the predicted entity and the known entity. Additionally, the BMKGC model incorporates a bi-encoder architecture, enabling simultaneous predictions at both the head and tail. Furthermore, we propose a straightforward technique to augment positive samples, mitigating the problem of degree bias present in knowledge graphs and thereby improving the model’s robustness. Experimental results conclusively demonstrate that BMKGC achieves state-of-the-art performance on the WN18RR dataset.

Zero-shot text rankers powered by recent LLMs achieve remarkable ranking performance by simply prompting. Existing prompts for pointwise LLM rankers mostly ask the model to choose from binary relevance labels like “Yes” and “No”. However, the lack of intermediate relevance label options may cause the LLM to provide noisy or biased answers for documents that are partially relevant to the query. We propose to incorporate fine-grained relevance labels into the prompt for LLM rankers, enabling them to better differentiate among documents with different levels of relevance to the query and thus derive a more accurate ranking. We study two variants of the prompt template, coupled with different numbers of relevance levels. Our experiments on 8 BEIR data sets show that adding fine-grained relevance labels significantly improves the performance of LLM rankers.

In this paper, we introduce the Financial-STS task, a financial domain-specific NLP task designed to measure the nuanced semantic similarity between pairs of financial narratives. These narratives originate from the financial statements of the same company but correspond to different periods, such as year-over-year comparisons. Measuring the subtle semantic differences between these paired narratives enables market stakeholders to gauge changes over time in the company’s financial and operational situations, which is critical for financial decision-making. We find that existing pretrained embedding models and LLM embeddings fall short in discerning these subtle financial narrative shifts. To address this gap, we propose an LLM-augmented pipeline specifically designed for the Financial-STS task. Evaluation on a human-annotated dataset demonstrates that our proposed method outperforms existing methods trained on classic STS tasks and generic LLM embeddings.

Text-to-SQL aims to convert natural language into structured query language, which is a challenging task. Current research focuses mainly on read operations and ignores other aspects of database operations such as create, update, and delete operations. The benchmark datasets as well as models that have been proposed also fail to cover these operations, limiting the development and practical applications in the field. To bridge this gap, we propose CRUDSQL, a large-scale cross-domain single-table CRUD operations Chinese Text-to-SQL dataset. The dataset contains 10,000 question/SQL pairs involving 625 tables from different domains. To support further research on this dataset, we also propose a baseline method, CRUDParser, which employs a two-phase approach based on BERT and T5 for SQL generation and incorporates two strategies, value matching, and value prompting, for interacting with databases to further improve the performance. The experimental results show that the new operation types bring different challenges for future research, and our approach achieves 67.08% and 83.8% exact set matching accuracy under both read and delete operations in the test set, but only 49.6% and 61.8% under create and update operations. We believe that the proposal of CRUDSQL as well as CRUDParser can provide new directions and possibilities for research and practical applications in the field of Text-to-SQL. The dataset is published at https://github.com/bizard-lab/CRUDSQL.

Large language models (LLMs) have shown remarkable adaptability to diverse tasks, by leveraging context prompts containing instructions, or minimal input-output examples. However, recent work revealed they also exhibit *label bias*—an undesirable preference toward predicting certain answers over others. Still, detecting and measuring this bias reliably and at scale has remained relatively unexplored. In this study, we evaluate different approaches to quantifying label bias in a model’s predictions, conducting a comprehensive investigation across 279 classification tasks and ten LLMs. Our investigation reveals substantial label bias in models both before and after debiasing attempts, as well as highlights the importance of outcomes-based evaluation metrics, which were not previously used in this regard. We further propose a novel label bias calibration method tailored for few-shot prompting, which outperforms recent calibration approaches for both improving performance and mitigating label bias. Our results emphasize that label bias in the predictions of LLMs remains a barrier to their reliability.